Spatial correlation of ground motions in estimating seismic hazards to civil infrastructure - Seismic Risk Analysis and Management of Civil Infrastructure Systems - page 72

Summarizing the conclusions and recommendations made, for example,
by Rhoades and McVerry (2001), Goda and Hong (2008b), and Sokolov
and Wenzel (2011b), the following scheme for consideration of ground-
motion correlation may be suggested: if information on correlation models
is not available, it is fi rst necessary to obtain upper and lower bound esti-
mates by assuming the extreme characteristics of correlations (the perfect
correlation: ρ η = 1.0; ρ ε ( Δ ) = 1.0; and the uncorrelated ground motion: ρ η =

0.0; 0 km). Second, if the difference between the estimates

based on the extreme bounds is signifi cant, one should make assumptions
about the correlation based on reported models and results. However, it is
necessary to bear in mind that the ground-motion correlation, in principle,
depends on the chosen ground-motion model and the within-earthquake
correlation structure may depend on local geology as well on azimuth-
dependent attenuation along the propagation path (e.g. Sokolov et al. , 2010,
2012). Thus, a single generalized model of spatial correlation may not be
adequate for some cases.

ρ ε ( Δ ) = 0.0 for Δ >

3.4 Future trends

The within-earthquake correlation depends on ground conditions of the
sites. Baker and Jayaram (2008) and Jayaram and Baker (2009) discussed
the infl uence of geotechnical characteristics (average shear-wave velocity
on the top 30 m of the soil, Vs30) on the correlation for short-period
ground-motion parameters. Sokolov et al. (2010, 2012) found a signifi cant
infl uence of geological conditions (both shallow and deep geology) on
characteristics of the within-earthquake correlation in Taiwan (see Section
3.2.5 and Fig. 3.2). However, no systematic research on quantifying the
dependence has been conducted so far. Thus, there is a necessity in intro-
ducing a model, which relates spatial correlation of ground motion param-
eters with characteristics of variation of geology in spatial extent. Such a
model will be very useful for loss estimation in the regions, where strong
motion data are scarce or not available; with the developed mode, spatial
correlation structure can be characterized by suffi cient geological and geo-
technical databases.
Another important application of the geology-correlation models is char-
acterization of uncertainty of Shake Maps (e.g. Wald et al. , 2008) using
conditional simulation of ground-motion fi eld, given observations from
recording stations (Park et al. , 2007; Crowley et al. , 2008b).
Obviously, development of correlation models for ground-motion param-
eters, other than PGA, PGV and response spectra that may be used for
regional loss estimation, such as seismic intensity and acceleration spectrum
intensity (Bradley, 2010), is a fruitful area for further study.